PROBLEM STATED I'm at a cross-roads in my space-exploration game design. I can either create a realistic (200 Billion + stars) galaxy where every time the player starts a new game, it's in the same universe, or I can make smaller globular clusters (~ 1 Million stars) of roughly 1000*1000*1000 LY, and have up to 2^16 different maps for the player to explore. I can see advantages and disadvantages to both. REALISTIC SPIRAL GALAXY One huge, realistic galaxy sounds attractive as a feature. I wouldn't have to worry much about dimensions and shapes, since I could base it off of our own galaxy. Also, there's plenty of room to have large uber-empires that stretch for thousands of light years and still have room to have nowhere land that doesn't belong to anyone. (Which Sol is in). Also, a player community could (hopefully) develop where players can share stories about planets and empires they found, and other players can explore the same empires in their own game. That's another attractive feature. Also, there's a nice psychological knowledge that this galaxy is the same for every game you'll play. Making the galaxy seem much more like it actually exists, and that it is old, and not just mushed together by teh processor 5 minutes ago. On the downside, I would probably have to have a 64 bit random number generator (takes 64 bit seeds. Actually I'd only need 38 bits, but that's beside the point) in order to allow for all the stars to be unique, which could slow down random number generation, which is a real bottleneck for some functions in game. And if the players got bored with the galaxy (or rather the stars you can access near Sol, since it takes alot of tech to reach the other side of the galaxy) they can't get a new galaxy, they're stuck with it. GLOBULAR CLUSTERS Globular clusters on the other hand allow for 32 bit seeded random number generators, which I already have several I've been working on. If the player got bored, there'd always be a new one to load. On the downside, the clusters would have to be realatively small. Meaning that there couldn't be any uber-empires. Alien empires probably couldn't get much larger than several hundred stars before they would begin crowding each other geographically, making the galaxy a land grab, which is not what I'm going for. Although there are a finite number of galaxy seeds, player's can't swap stories as easily, since most of them won't be on the same universe. And the psychological idea that the universe exists is gone, since you can just make a new one whenever you feel like. POSSIBLE SOLUTIONS 1. I could have Sol's location in the huge galaxy be player defined. So you could start a game in a different part of the galaxy each game if you wanted, which solves the problem of novelty. 2. I could make both. But until I finish one version, I won't know how much work this involves. It could be nothing short of impossible. There are other solutions, I'm sure, but I can't think of them. CONCLUSION Which would you prefer as a player. The problem in either case isn't running out of stars to explore (or even overhead for the computer), but about novelty at the start of each new game.

Quote:Original post by Numsgil
POSSIBLE SOLUTIONS
1. I could have Sol's location in the huge galaxy be player defined. So you could start a game in a different part of the galaxy each game if you wanted, which solves the problem of novelty.

This would provide more than enough novelty for starting positions, so if this fits in better with your overall concept I'd go with this.

Of course, if space travel is relatively easy, then even if you always start at the same point, your path of exploration will have many opportunities for divergence. So I'm not even sure that that novelty is even needed.

I honestly don't think the payoff will come from varying the stars, and there are many ways to make the universe feel large and alive without killing yourself. Refering to the galaxy visually (using powerful imagery that locates them in the galaxy) and contextually (names like Alpha Centauri or Procyon that they've probably heard of) I believe can allow you to create a galaxy that is meaningfully large.

It won't be too interesting if the galaxy has large pockets of undifferentiated mass. It will be like flying over a forest and seeing tree after tree until you get sick of the sameness. With a huge galaxy you run into the challenge of making areas special and memorable. There are an awful lot of red dwarfs out there, for instance, which are just there, with nothing special about them.

Size can be illustrated with different approaches, you don't neccessarily need 10~billion stars to show you how huge the galaxy is. Some examples of this:

The ElderScrolls 2: Daggerfall, had probably a couple hundred thousand towns, villages, and cities all over the map, god they had lots. But were they different? Not in any meaningful way anyone could see, they were all pretty much cookie cutter's, which is the problem you may inevitably run into without alot of diverse content to fill that void of space. I suppose they wanted to illustrate how huge the world actually was, but ended up boring the crap out of the player since there was no difference between going to the far off reaches of the island, as opposed to staying right where you were.

On the otherhand, Morrowind managed to create a very large and diverse world over the same equivalent land-mass size that Daggerfall used. The size of the villages and content was considerably larger and much more detailed. It takes quite some time to run from one end to the other, with many memorable hardships and sights along the way, dispite having a fraction of the number of villages Daggerfall had. Alot of detailed content can help make a world look and feel alot bigger than its scale would suggest, and is probably more memorable and enjoyable than exploring a billion look-alike towns.

Here's a data point: Does the user have to click on a star and do something specific to explore each star, or can he set autonomous bots to find new stars? If there's clicking involved per star, then exploring even one percent of a million stars would require 10,000 mouse clicks.

One million stars is a HUGE number. You could have 100 players play in a single million-star galaxy and have most of them never run into each other, unless you allow very rapid, semi-automatic expansion, and provide incentives for players to actually meet each other.

1,000,000 stars divided by 100 players means 10,000 stars per player. What is the management interface in the game that allows the player to manage that many stars in a reasonable way?

I'm not sure about your actual game, but in any sort of 4x style game, a random map generator is the #1 priority for me. MOO3 even kept me playing for about a month with a passable RMG.

I'll try to address all the concerns posted here, so bear with me if this is a long post.

Quote:
This would provide more than enough novelty for starting positions, so if this fits in better with your overall concept I'd go with this.

Of course, if space travel is relatively easy, then even if you always start at the same point, your path of exploration will have many opportunities for divergence. So I'm not even sure that that novelty is even needed.

I actually really like the idea of defined start positions the best, so I think I'll go with it. You can start at different spots, which will change the evolution of your game considerably.

Space travel is difficult at the start of the game, and becomes infintely easy near the end. So it becomes very important to have starting stars that are of interest to the player. I think having a new start location every game solves this problem nicely.

Quote:
I think I'd have to know more about what you do in this game to answer.

Sounds interesting, especially since I'd like to do something similar. For mine, I need the large universe because I'd want people to share stories, but I think you can do the same thing for the smaller maps, provided each 'story' also provides what seeds are used, but then people would have to quit playing their map. Although why can't you link the smaller maps together?

I apologize for not telling you what kind of game this is. I've posted alot of threads since about July onward, and they've all been dealing with this game, so if anyone is interested, you can look them up.

In a nutshell, it's a 4x empire building game with no empire and no rivals. You play until your character dies of combat or old age, and the emphasis is on exploration.

As for linking smaller maps together, that's really how the larger galaxy works anyway, a bunch of sectors linked together.

Quote:
It won't be too interesting if the galaxy has large pockets of undifferentiated mass. It will be like flying over a forest and seeing tree after tree until you get sick of the sameness. With a huge galaxy you run into the challenge of making areas special and memorable. There are an awful lot of red dwarfs out there, for instance, which are just there, with nothing special about them.

Yes, there is a problem in making each place more or less unique. Other than having different land masses, etc. planets of similar types are, well, similar. I think this is probably how the real galaxy works too.

After all, I'm using fractal techniques for generating most of the planets, and the idea behind fractals is that they're self-similar.

The way the generation of planets works, it is just as easy for me (and the computer) to have 200Billion as it is to have 100. Meaningful interaction with the expansive world will come (I hope) from the interaction with the player.

It sounds fun in my mind, and I don't know that I can translate it well. If the game turns out to be boring, well, I guess this will be a laernign experience.

Another example of an expansive universe is in Noctis. His game runs in DOS and has 70 Billion stars. There is still an active player community, even after several years. And his game is more of a toy than a game, since all you can do is go to planets and explore.

Quote:
Here's a data point: Does the user have to click on a star and do something specific to explore each star, or can he set autonomous bots to find new stars? If there's clicking involved per star, then exploring even one percent of a million stars would require 10,000 mouse clicks.

One million stars is a HUGE number. You could have 100 players play in a single million-star galaxy and have most of them never run into each other, unless you allow very rapid, semi-automatic expansion, and provide incentives for players to actually meet each other.

1,000,000 stars divided by 100 players means 10,000 stars per player. What is the management interface in the game that allows the player to manage that many stars in a reasonable way?

Everytime the player reaches a new star, he can map any new stars that are around him. The vast majority probably are of no interest, so the player will ignore them. But maybe he finds 10 new stars worth exploring.

Basically, there are alot of stars that are little more than pretty.

What I really like is the size of the galaxy, more than the number of stars. I want the player to travel thousands of light years, and meet empires that are ten thousand light years across.

There is a semi-automatic colinization mode, but its use is more for filling in the gaps than active exploration.

Quote:
but leave the space open for the player to see... just make alot of random generated differences... like most star systems should have atleast a station or an asteroid belt, said station should have services maybe even some odd mission

The beauty of the random generated galaxy is that each of teh 200 BIllion stars can have up to 10 planets, and each of those planets can have up to 30 moons, and each of these bodies can be landed on and can (hopefully, if I can master OpenGL or Direct3D) be explored first person.

I know enough to know it's possible. The diversity is more of an issue of how much time I have.

Not all systems have anything in them, as far as intelligence. The player should probably spend his time exploring intelligently. Alot of information on each system is available before teh player explores it, including any man-made radio signals, etc.

Quote:
I'm not sure about your actual game, but in any sort of 4x style game, a random map generator is the #1 priority for me. MOO3 even kept me playing for about a month with a passable RMG.

It's a 4x empire game where the player doesn't deal with the logistics of running an empire. But even more so than in 4x games, the map generator is important.

To give you an idea of how the game works:

Master seed->Sector Seed->Star Seed->Planet Seed->Moon Seed

So when a player arrives in a new area, all the elements of the area are constructed from the seed, which is itself easily obtainable from the master seed.

Say a player arrives around a planet. From the planet seed, a terrain heightmap and biome map are created. If teh planet is inhabited by aliens, a map of cities are made to. If the player arrives inside a city, the location of the buildings etc. are decided by the city's seed.

Many cities will look similar, since it's hard to generate new graphics, but they all are different in layout, and different alien races have different personalities which affect the contruction of their cities, etc.

So having 100 or 200Billion stars doesn't increase my complexity in programming at all. I imagine most of us would prefer 200 Billion to 100 if there are no increases in overhead, both as players and programmers.

However, having a galaxy which spans 100 000 lightyears from end to end allows for some 'epic' elements that a smaller galaxy doesn't allow.

Oy. And I thought MOO3 was bad.

More than a million stars is what I personally would call crazy. No player will ever be able to explore them all. Even so, I think that's probably the maximum you could get away with for a high-level 4X game.

What you suggest would be really, really more suitable to a single sector of space or something. I'd put a max of maybe 10,000 to keep it playable (unless this is something designed for, like, 100+ players).